Malignant gliomas are common primary human brain tumors, but problems in their pathological classification compromise patient management. These difficulties have sparked considerable interest in molecular genetic approaches. Clinically relevant genetic associations have been discovered, but practical problems have prevented widespread diagnostic application of this knowledge. We hypothesize that custom array comparative genomic hybridization (aCGH) could provide a sensitive, specific, cost-effective and rapid method to assess human malignant gliomas for a variety of clinicopathologically relevant genetic changes and that such first-generation custom CGH arrays will provide the basis for improving diagnostic correlations for future assays. To evaluate this possibility, we propose a two-stage plan that capitalizes on our existing strengths in molecular genetics, pathology, biostatistics and clinical databases. For the R21 component, we will: 1) evaluate custom aCGH sensitivity and specificity in comparison to standard assays; and 2) generate custom BAG arrays for CGH that include targets providing broad genomic coverage as well as focused coverage of chromosomes 1, 7, 9, 10, 19 and X, and other select loci. Once we have met the Milestones from the above R21 Aims, we will proceed in the R33 component to: 1) evaluate whether alterations of particular regions on the assayed chromosomes, as revealed by aCGH, offer improved and/or novel correlations with chemoresponse and survival in two carefully annotated cohorts of malignant glioma patients: a) a retrospective series of anaplastic oligodendroglioma patients; and b) a prospective series of malignant glioma patients. We will then: 2) develop an aCGH-based classification of malignant gliomas using the data from Aim 1 of the R33. The long-term goal of this project is the implementation of a practical molecular assay to detect a variety of clinicopathologically relevant genetic alterations in malignant gliomas. We also anticipate that such information will contribute to identification of key glioma genes as well as to construction of next-generation diagnostic approaches. Given these endpoints, the application is highly responsive to PA-04-102, "Phased application awards in cancer prognosis and prediction."